py-why / graphs Goto Github PK

We need a good way of implementing the "Views" for output of various MixedEdgeGraph functions.

Is your feature request related to a problem? Please describe.
A clear and concise description of what the problem is. Ex. I'm always frustrated when [...]

The current implementation of m-separation relies on transforming the existing graph into a directed graph with hidden variables standing in for the bidirected edges, and then calling networkx.algorithms.d_separation. As pointed out in the function comment this is not efficient for graphs with many bidirected edges, and in the worst case might be O(2n).

Describe the solution you'd like
Improve the implementation by adapting the current networkx.algorithms.d_separation property directly. This should result in an O(n) solution.

Describe alternatives you've considered
I considered adapting ananke's m-separation property but realized that implementation is actually O(n^2) which is worse than the O(2n) solution here.

Additional context
Add any other context or screenshots about the feature request here.